ML20091M864
| ML20091M864 | |
| Person / Time | |
|---|---|
| Site: | Byron, Braidwood  |
| Issue date: | 08/25/1995 |
| From: | MPR ASSOCIATES, INC. |
| To: | |
| Shared Package | |
| ML20091M806 | List: |
| References | |
| PROC-950825, NUDOCS 9508300320 | |
| Download: ML20091M864 (48) | |
Text
{{#Wiki_filter:, _,_.__s.-.____.__.,..,.-.-_.-_....~._.-.
,e .
jd s 1
. o
'.f f* - l
- , s .
S , . ,g '
,2. MP~3 A8888888 - -
> a LAPPEND11 B m ,
z - f:, p y.
, }'i s
-i
;i:!
i s , , ,
.t.
3 , , - 3 j
.i i .4
- o. r .
.. g
/ '
TVerificatihn Analyses for. Cosaputer. Program'
, : 3 -a.e a
1
'9 I
t i o >b 4 t j
- j. _.. i j ". -i r
e 'f 1 ( :'s ' (=.- 1 !~ k u. 4 *
+
! .- i I. i L, i n 1 e,. . . 1- 1 r: :p + ts , :n , I'.- g ,f e , l 4 s j i* s , $ - t. A '. . e.. o 6,T .t. 4 i : o' i 4 u :. t ,.. . . . = , 4 /9508300320:950825 ~ P I y"f m. .: PDR: ADOCK.05000454- 'i e., JE i..
.. a; .s, *
. - . . _ . . . . t. .m P.DRL-- .
.._Z....-
_ . - - , . - - . _ , . . - . _ _ _ . . . . . _ . . - ~ . . , . .
c l f B.1 Introduction To check the results obtained from the transient thermal. hydraulic computer program, several test problems which have' analytic solutions or allow comparison to experimental data were run using the program. These test problems were chosen to check \ 1 those features of the program which were considered _ auxiliary feed pump important for the Important parameters for this turbine piping analysis. analysis were considered to be as follows: 1. The conduction of heat to the pipe wall from the l fluid which is responsible for the condensation of i steam.
- 2. The heat transfer coefficient between the pipe wall and the fluid. This coefficient includes regions of free convection, forced convection, and a
condensation heat transfer. 4
- 3. The slip model which determines the rata at which separation of water and steam occur in the piping. The time at which the water accumulates i
i B.2
- - ~- _
~
ct th3 inlSt to tho turbinDo 10 dep;ndsnt on thic model although the total water available would probably not be very sensitive to this parameter.
- 4. The conservation of mass, momentum, and energy as <
solved by the program including the appropriate equations of state. i B.2 Comparison to Analytic Solution of Heat Conduction
' Throuch a Cylindrical Wall _ ,
Figure B1 shcws the computer model which was used to , obtain the solution for the transient conduction heat t transfer through a cylindrical wall similar to that auxiliary feed pump which exists in the _ __ i turbine piping. The pipe wall was assumed to be at an initial uniform temperature of 100 degrees Fahrenheit 0.2 and a constant heat transfer coefficient of r Btu /sec/sq f t/deg F was assumed for the inside surface of the wall with the fluid temperature at 500 degrees 4 F ahrenheit. The outside wall of the pipe was considered perfectly insulated. The solution obtained assuming using the thermal hydraulic computer program, several different numbers of elements through the pipe wall, is compared to an analytic solution for this < problem in Figure B2.- The comparison indicates that the thermal hydraulic computer program provides B.3 1
. 1 z .
\
i l racconnble antwara to thic probica and that tha pipa ! wall can be adequately modeled by four elements across its thickness. l 1 i B.3 Reasonableness of Heat Transfer Coefficients The thermal hydraulic computer program used for this analysis provides as part of the print-out the equivalent heat transfer coefficient for each heat flow ; path.. The~ magnitude of the heat transfer coefficients were checked for reasonableness against hand calculations and published results. B.4 Comparison to Test Data From Reference 3 Reference 3 provides measured results of void fraction and vessel pressure for a test vessel ~which was partially filled with saturated water at high pressure and then blown down through an orifice. The configuration of the test vessel is shown in figure 83 as it appears in Reference 3. As the vessel depressurizes, the saturated water inside the vessel flashes into steam. The void fraction as a function of , l height in the vessel at different pressures during the blowdown provides a good test for the effect of slip between water and steam as the steam tries to escape from the vessel. Figure B4 compares the measured void B.4 1
fraction ac -provided in Rotoranco 3 ct 10 socrnda end at 40 seconds after initiation of the blowdown to those which were calculated by the thermal hydraulic computer l l program used to analyze the ,_, ,_ auxiliary feed pump steam supply piping. The comparison shows that the slip model in the computer program is adequate to determine qualitatively the separation of steam and , water which would have occurred during the June 9th 1 transient at _. Figure B5 compares the rate of depressurization from Reference 3 to that calculated by the thermal hydraulic program used to analyze the _ ._
._ Auxiliary Feed Pump Turbine steam supply piping. The ccmparison indicates that a slightly more rapid depressurization is calculated than was measured; howe ve r, for both the pressure rate of change and the void fraction ,
distribution, the agreement with the measured data is considered good. The lower calculated value for the ! pressure at 40 seconds is consistent with the higher
- void fraction which is calculated for that time.
G I B.5 (
U.5 Analysin of a Plo7 Comprancion Wnve l As a final check on results obtained from the therma hydraulic computer program, a long pipe at high pressure and closed at one end was analyzed for the effect of a sudden increase in pressure at the open boundary. Superheated steam was used as the medium The analytic solution of this problem i.t side the pipe. requires-that the shock front travel down thedp,ipe end at l the speed of sound and is reflected from the c ose The speed of sound is at twice its original magnitude. t not an input to the thermal hydraulic program and i mus
' be obtained from the solution of the fluid d by conserva
' equations combined with the equations of state use j
Consequently, the solution to this the program. id problem is considered to be a good test of the flu i ^ dynamics as they are solved for the Davis-Besse June 9th transient particularly since opening the steam l __ auxiliary feed pump admission valves in the _,, ,_ steam h piping (which admits high pressure steam from t e i generators into the low pressure region of the h f auxiliary feed pump steam piping to pressurize t e piping) is similar in nature to the sudden l d for preswurization of one end of the pipe as ana yze this problem. The configuration used for the test problem is shown in Figure B6. I i B.6 i l
Rsculto from the tect problem cra chown in Figuro B7 These results indicate that the -wave front which forms due to the sudden pressurization holds its shape and travels at a speed of about 1950 feet per second once it has diffused over a few control voinmes. The speed 3 of sound for superheated steam at 500 psia and 600 degrees Fahrenheit is given as 1860 feet per second in Reference 4. The reflected wave from the closed and of I the pipe also behaves as expected. Consequently, the effect of the much slower pressurization which occurs , at , when the valves in the feed pump turbine i steam supply piping are opened on a slow ramp in time should be determined adequately by the thermal hydraulic computer program used for this analysis. ' i e i f B.7 4
'Y
,r i
l
- 0.22" Pipe Wall-l Outside Insulated
. Initial' Wall
- Temperature 100 Degrees Fahrenheit- , p n a +
l 0.2 Stu/sec/ft 2 /dag F j Beat Transfer Coefficient ' with 500 degree l T2.uid Temperature i . 6"_O. D. , ! _ ~ Plag,Qgn @ g, b l l 0 r 4 1 i Conductivity: .0022 Stu/sec/ft2 /deg F Heat Capacity: .12 Btu /lb/deg F l Density: 490 lbs/ft 3 l l l i FIGURE B1 Schematic of Pipe Used for Thermal Conduction Problem I l l l
600 Analytic
- 20 Elements 500
- a 4 Eleznents 400 -
Solution at [ 5 seconds 300 Solution at 200 - 100 - Initial Condition 0 - - - 0.0 0.25 0.5 0.75 1.0 Fraction of Pipe Wall FIGURE B2 Solution for Heat Conduction Through Pipe Wal3.
computer-Model . Test configuratica . Blowdown 14 A 15 u Orifice i 13 noundary 8 0 at l 12 Atmospheric 11-Conditions 10 9
=:_-.
8 14 feet ~ 7 i 6 1 Blowdown 5 ;
-+ 1 foot +- Valve ,j F 4 q ,
_ _ - - _ _ _ - - - 3
~
Suppreseior - - 2
~
~
1 4 FIGURE B3 d' Test Configuration from Reference 3 and - ' Computer Model c _ . _ _ . _ _ _ _ _ . ____ _ _ _ _ _ _ _ _ _ _ _ . _ _ _ _ _ _ _ _ _ _ _ _ - - . - , - . . . . . . - 4. ._ * - - . . . - , , - . , , - - , , . m
.At 40 Seconds'
~At 10. Seconds ss-1.0 -
.a Measured
* ~ "
.p ' Measured x Calculated p . Calculated
.8-
.8 .
6 -- .
. void -
Fraction .6 - l D = i
.4 -
O r
.4 -
y p [ y , s ~
,0* O
- De v i U" # .2 .
- D I
.2 -
,0* a :
Y F~ 0 ; e . t i i i 6 8 10 12 14 0 e i i 0 2. 4 8 10 12 14 6 0 2 4 Axial Distance in feet i Axial Distance in feet frosa Bottom . from Bottom A FIGURE B4 . Comparison of Calculated and Measured Voida for Test from Reference 3
l P 1000 . 900 ,,
- Measured 800 - 3 Calculated 00 -
Pressure (PSI) 600 ,, 500 , 3 400 . J J 300 . g l 200 . L 100 -
' t 0 300 l 150 200 250 0 50 100 I - -
Time (seconds) i r i ^ t ' FIGURE B5 Comparison between Measured and Calculated Pressure ; for Test from Reference 3 i i l 2
2 feet Pipe Diameter: 500 psia Initial Pressure: 6000 F Initial Temperature: Closed at End Prassure Boundary at 600 paia 100 feet D FIGURE B6 Geometry Used for Pressurized Pipe Test Problem t
a
- - 800
' t I 800 Initial conditions Pipe are 500 psia at .0593 seconds and 6000P superhea(
steam
- 600 600 - -
/ [
Pressure (paia) at .0193 seconds at 0393 - 400 400 , seconds at 0073 seconds
- 200 200 -
_0 s 0 _,l I i 100 60 80 20 40 0 i FIGURE B7- t Pressurefor Pipe Distribution Pressurized at Various at End Times ,
' Enclosme 3 Paper Presented to ASMPJFRK Power Generadon Conference i
I l 4 i
^""" "** --#
In the connector represesting the pump is modified - ! accirdingly, le the second case, for example after a N pump trip occurs. the esestdeun of the p e p depends en the interaction of the peep with the fluid and a K\\\\ / dynamic torque belance en the pump shaft. -In the e ' second case. the seletion of egustions in addition to. - +
. the fluid dynamic equations will be reeutred to /
F "'" , , represent the peep behavior. The rate of change of \ , , tile speed of the pump is given by the terque divided by the retary inertia of he pung shaft assembly. ' 4 t (12) w= --- .
*e I
! = where: t :is the tiet torque on the pump 1 is the rotary taertta of the pump shaft assembly l
' The not torque on the pump Includes the torque due to the pump motor, the torgue app 11ed by the fluid, torque from losses in the bearings, and torque due te . \ f
' losses from the fluid. . aus (13) i t.= t,- tr th-tf1 f!GURE 2 where: t, is the torque applied by the mocor Typical Check Valve Geometry
} W tf ' = AP p is the fluid torque e is the angle of the disk I where: Prof t is the not torque on the disk assembly ' w I J tb "Ib-"' ref (88-I cef 1 is the rotary inertia of the disk The net torque tending to rotate the disk includes i W the fluid forces, the buoyant weight of the disk. j 1f1 = ( 1 1 - i b) Pref ( - )2 friction in the shaft, and spring load if the check
'd' ref Wref valve has a spring.
is the pressure difference across puer t = tf + t, 4 tb + ts (III , SP is the mass flow rate through penP W is the power at full speed Prof is a bearing loss coefficient where: if= AP A 0R D is the fluid torque on the disk l fb is the pump efficiency sin ( * . e fh) is the torque des to
- q tw= W RD eg weight of the disk
.2 IAhta Falves are included as flow arets which change as a NDer is the torque due to friction t b"#'R %cg $ en shaft from centrifugel f function of time. The perameters in the annentum .
equation which are affected incluce the inertial lead on the disk length and the flow loss coefficient. Reducing the ' area of a connector will increase the pressure drop M is the friction coefficient en the shaft ; across the connector and will cause the flew rate ts- KgR3 sin ( de - 4 s) isspring the torque due to a ' threagh the connector te decrease. Imposing a change in a flow area as a fumetton of time en a flow 4' connector is strat htforward and is easily handled Ap is the area of the disk
- directly in the so ution for the (fuld equations. :
F similar to the imposition of a time dependent pump RO is the moment are from the shaft to the speed on the fluid solution as discessed previously. center of the disk However, in the case of a check valve, dynamic r equattens which determine the positten of the check WD . is the benyant weight of disk assembly valve disk based on a force balance on the disk pust - 1 be added to the fluid solution in a menner similar te is the mass of the disk assembly MD the solution required for the pump speed when the ; speed is affected by the fluid conditions. A typical Reg is the sement are from the shaft to the l
' swing check valve geometry is shoue in figJre 2 and center of gravity of the disk asseadply the equation of action for the disk is given below,
[ 4 efh is the angle at which disk hangs freely. l t-l
.. (14) l q R$ is the radius of the shaft
= -
1-4 4 e ? s.
X, is th3 spring co:stant TJ si the tesquereture at node j Rg is the soment are from the shaft to the spring Rg is the thermal resistance for connector k I e s is the angle at which the disk contacts gt is t.be total heat flow in connector k the spring
'* AHk is the heat transfer area of connector k rwuetian Hear Trander j {g A conservation of heat flux approach is used to solve The integrated form of Equatten (16) becomes:
for the transient temperature distribution in the i( metal parts of the geometry by dividing that geometry Into a nunper of control volumes and reesiring that T.,,3- 7,
+ Mt(Tr TjuRg - 5,Voi n (21) the not heating rate in each volume satisfy the PCYol a at ,1 . energy conservation eg'Jation. The not heat rate for on n a J
each volume is written as a function cf the temperature at the centroid of that volisse and the where: Se is the heat generation per unit volume in resulting al etrait equations arr solved luglicitly. volume a The heat to uction equatten e n be written in the following form. Volg is the volume of control volme a T 18 the temperature in volume a at the 1
' OC ,) g
= V. J s s (II) beginning of the time step nt To ! ic the temperature in volume e at the I J . .k 7T (II) and of tne time step where: J is the thennal heat flux There vlT1 he one equation like Eesation (21) for 5 is the heat source per unit volume each volume in the structure. The total set of k is the thermal conductivity equations can be represented by a single matrix T is the tesperature equation of the form:
< fe is the density C is the specific heat capacity A T = B (22) t is time j where: A is the matrix of coefficients in Eq. (21) Equation (II) represents the condition that the rate T is the vector of unknown temperatures i
- of change of temperature at a point depends on the B is the right hand side vector j rate of heat flow to that point and the rate of heat
; per. oration at that point. Equation (17)isthe Boundary conditlops must be imposed on Equation (22) definition of the heat flux, which is proportional to to represent the effects of boundary conditions on the temperature gradient and flows from higher the geumatry being analyzed. Three different types I temperature to lower tosperatura. The conductivity of beundary conditions are provided. These boundary is just the preportionality constant between heat conditions are treated as surface connectors which -
flux and the temperature gradient. By substituting contribute to Equation (22) as additional heat flows Equation (17) into Equation (16) and assuming the as folloes. thermal ccaductivity is not dependent on position. l the norsel steady state heet condJction is obtained. 1. Temperature specified on boundary. The i additional heat flom represents a resistance . 4T 2 path between the pode at the centroid of volume ra and the surface which is at a specified PC4t- + k7T = 5 (18) temperature. TB. { The numerical approach used to solve the heat Qk"AI(Im-II)/RIk t (23) conduction equation is based on Equations (16) and where: A$g is the heat transfer area of the path i (17) directly rather than on the heat conduction ! equation given in Equation (18). By dividing the region of solution into a la e number of volumes R5k is the resistance between the node which are connected to each o her by heat flow paths tha valine centroid and the surface [ er connectors, the method integrates Equation (16) i over each volume and describes the heat flow in each 2. Heat flux specified on boundary. The additional j canaector in terms of the Lamperature at the center heat flow is specified directly by the heat flux of the connected volumes using the following on the beundary, JO. numerical approximation for Equation (11). Ok . A$k J3 (24) Jk = l Tg. Tj) / Rk (19) Heat transfer coefficient specified on boundary. 3. Og - Allg Jk (20) The additional heat flew is represented by the resistance path between the center of volume a where: Jk is the heat flux in connector k between to the surface which is at temperature TB with a modes 1.j heat transfer coefficient, liTC. i l 5 3 E
- m. . . _ . . . _ -~ __ _ . _ _ _ _ _ _ _ _ _ _ _ _ _ _ .
fluid solution. In the second cooperison, the f
/ 1"I,'$"% a g suam smeen transiest heat conduction was calculated for a pipe wall giren a specific heat tressfer coefficient and I
3/!g8,;,,",,l"*l,* Lag,% an 1:ternal fltid toeperature. The second casa' tests 5e==== sa.- the memorical approach used for the heat conducties soluttee. ; we es+=dsw For the first case, the analytte solution predicts l Nu asser r n'e"j the estabitshment of a were front at the end of the pipe which is suddenly pressurtzed and the wave front na se rs en propagates along the pipe at the speed of sound, on uneet ser ens no w reflecting from the closed and of the pipe at twice om , ca s., the incoming ampiitude. The poemetry seed Ia the analysis and a descriptise of the reselts obtained e e n '* **'*a a
- are provided la Figure 3. The results indicate that ase u
the speed of proeagation of the wave front obtained
""'8"'"*' % ,--- from the numerical technique agrees well with that f ~, published in heference 2. The propagation velocity
., of a pressure pulse obtained from the numerisal 3
A 3 k { soluties is only a function of the equations .of state ge "- g* g*L;*#,8 m-y and the conservation equattens employed, and .is not espitettly provided to the solution techniese.
,, ,,,,, k;T ,
"*" The poemetry and a comparison of the results a$btained ;
o for the heat conduction problem using the numerical ! am, '" technieue to as analytic soluties are shown in ' Figure 4. In this case, the numerical technique provides accurate results for the temperaters
- . . distributten across the pipe 1 with as few as fear 8 "
" " weluees provided across the pi wall thickness, e r ciaew
'*" CtElpAR150M TO TURg11IE Trip l FIGURE 3 Test Problem for propagetien of a Pressure The approach outlined above was used to calculate the
- Ifare Treat in a Pipe transient pressure response for a turbine trip at a j l
nuclear power plant. 'The turbine interceptor valves l J used to isolate the turtise from steam flew are fast , acting valves which : lose in a fractica of a second Og - A5k (T, TB) / (1/HTC + RSk) (25) and can set up a rapid pressure increase in the i piping upstream of the valve. 4 The specified parts of Eeustion (11) are transferred Figure 5 shows a schematic of the system which was ts the 8 vector side of the equation and the unknown analyzed. The stesa generater, which supplies steam temperatures are obtained by solving the resulting to the turbine, will react to the pressure waves i which propagate through the pipe and the complete
- metris equation.
< steam generator was included to the model. The heat
- DESCRIPTION OF C0frUTER PROGRAM transfer u the steam generator aise affects the response and was included in the model. la the=
The analytic approach described above has been particular case analyzed here, pressure measurements l in the pfplag were not available; however. the incorporated into a computer progree to sinyllfy the evaluation of pipe nation during transients. A absolute pressure in the steen generater and the i considerable nusber of pre-and Post processing water level which is based on pressure differenfial features have base developed to allow input data to asesurements across a part of the steam generator be entered eastly and to allow the results to be were recorded. The cosyseisen of calculated reselts interpreted quickly. The solution technique used has to measured results for the pressure in the stems i been optisited over the last several years to generator is shown in Figure 6 and indicates good ' increase the efficiency of the calculations to the agreement. The small pressure pulses recorded in the ! point where most transients which have been analyzed staae generator are a response to the larger pressure l vill run within a few hours on an 191 PC computer or swings which occur in 3e piping between the steam l generator and the intercepter volves. Calculated 1 compatible. 5taple tramstents will run in a few cLinutes. pressure at ties interceptor valve is included in _ ~ l
' figure 6.
l CCBFAA!50f TO ANALTTIC RI5ULTS The recorded water level in the steam generator is I Results obtained using the approach outlined above shown in Figure 7. This measurenset actually 2 were tempered to analytte solutions for two slaple esasures the difference in pressurg between two geometries. The first af these was a sudden points in the steam generator and consecuently congressise in a pipe containing pressurized steam, reflects the pressure oscillati,ons prvduced in the
. similar to what might occer igstream of a suddenly piping. Also shown in the figure is the calculated 2 closed valve. This problem tests the propagation mass flow rate leaving the steam generater. The characteristics of the numerical approach for the sensured water lerel and the calculated mass flew 6
l l.
' ^ ~ ~ , '"
w W.228 Pipe me11 let ] _ matrsia e 20 timmem48
. ee lege14ted . . u ..
tsea we&L me f , Ea,Se - - 8.2 I ste/ses/ft / des F [ pese Tsametes onessamaans w&te See e. gees i l Famid temperatuse
' 4De setussam et i l 9 secumes 4' p. 9. '
aos
, = .P h 4 6 st _ ~
'~~'
i see
/~ seletten at .
l ggg . _ Enttia! 1 coneAuse 2 Ceedoes&vitys .0033 954/seetsg fgeg r --
, e..t ce, sty, .1: swswee, I r g __
e.. e.n 1.s se.satts 499 16s/f6 ..e e.n r,..us. .: noe a w uv i F181mE 4 's9,, I Test Probles for Heat conduction Through a Pipe Wall l" -- f
' t
- y b =t toe 1 t __
l . F 7 t __ l $1'se.stees% sete as fee T_ f l M- W feet of min steam ) F , ]5 j N Lise
+
t Laus r .d w I I - m 4 _r
^
/ I b
f g-j seu r e -
$' l
,. 'L,,
1 r L~ h m ve yme e.s 4 g l
! I Y y nets f d C#
K' , f_ I .- I - i I._ i ,1l - 1 - a -
..m T _
i- -
; I .._
c 1 i tas 7 , w esli I ansessa.c 8'**=d"* 5H* N t
*e**'" #
FIGURE 5 primwr Camputer Model for Turbine Trip liet
) ,
i 1 I 1 1 a
The resses rete respond at atest the same frequency. the steen gamerater flow resat:s high at the end of the transient is encaisse safety valves and turbine saamma
.-. h. s. n to. .i 8 wl by 7 ass valves opened as a result of the transient.
[ j ! {'"' [
, COMPAR150el TO PIPf, IWTION DATA
} . "8 Transient data for pipe motion at power plants is 8*- difficult to obtain since measurements are not
, " normally avellable for severs treestents which cause However, the magnitude of s
u.s significant pipe metten. the mettee can sometimes be inferred free
" consegeestial damage, such as deformation of piping n.sy sepports er from scratch marks on piping or terti p g., y insulatten. Several plant transients have been
- , g nJ -
analyzed by the approach described above and results l pies. consistent with the observed damage have been l/ se.:y 4 4 imm' 4 obtained. Two such incidents whica have been analyzed are described below. j 'n' '
#dj j J* Two recirculation pumps are aligned in parallel.
"'ji [ , ,
",, l.
in order to allow the pumps to be operated individually, check valves are installed at the l
,, s t, discharge of each pisus to prevent the flow from
, .a a single running pump from passing titrough the
; .N secured pump. When both pumps are running and a.: one of them is tripped, the check volve at the ves , .
- 8 3 s a a e s a e se a u discharge of the tripped mano willcase In the closenatlysed, and a ,
g3 waterhammer event occurs. l significant deformation of a pipe support near the suction side of the tripped pump occurred. FICURE E An analysis of the transient was used to Pressure Comparison for Tuttine Trip accurately predict the deformation of the hanger une and to determine if the recesign of the htnger , i was adequate. as g,,,, "
- 2. A disk in a control valve broke frwe its 3" attachment suoport end freely floated within the Tne design of the valve was such f ' '" } talve body.
! that the disk was unstable in this condition.
I sn l
- j ,
" The interaction of the disk motion and the resulting pipe response set up a severe i
, 8" ' vibratten transient in the piping system, 1
8 dameging several supports on the piping system.
) 4
, , An analysis of the transient was used to r,btain e- i 8' ,. d i piping loads which were consistent with the e
- se, observed damage and the resulting calculsud i une l
stresses in the pipe were evaluated to
- cainated n* demonstrate the integrity of the piping.
1 ' ' #w CONCLUSJDN isse . The analytic approach described above, and its 4 laplementation in a couputer program which is easy to
, ~ use and runs on a personal computer, has proven to be m 't
)
, m -{ a useful and practical tool in design work and in
.o _
- evaluating power plant problems.
j j
j R(FERENCES J. H., Redfleid, J. A..
f 1. Porsching. T. A., Murphy,4; a Fully Implicit I and Davis. V. C., " Flash. t '
' .in Fertran IV Program for the Digital $leulation of
. i e Transients in a Reactor plant March IHf.
f i g ta= 65. -*i WAPD-TM 840, Bettis Atomic Power taboratory. FIGURE 7
. . Ceaparison of Steam Generator flow to 2. C. A. Meyer. R. 6. McClintock. G. J. Silvestri, Meawreft Water Level for Turbine Trip and R. C. Spenser. Jr., 'ASME $ team Tables. Fif th Edition' 8
3
I TRANSIENT TWO-PHASE recua,cu co r, BLOWDOWN PREDICTIONS OF AN INITIALLY STAGNANT SATURATED LIQUID STEAM IN A VESSEL USING TRAC-PF1 YASSIN A. HASSAN The Babmet a inicos Company, Nuclear Power Division P.O. Box 10933, Lynchburg, Wginia 24$06 0933 I Received October 16. 1984 Accepted for Publication January 14,1985 o Mf5MENNsTrEIEES$fNTMMEEMOS @- Comportsons of the predktlats of the best estimate pret- { ,L surized water reactor TRAC-PFI/ MODI computer code to l esowsnm data of the General Electric levelswelltests wre performed.
- l -
Ortf\ce Various Isme-step sites and nodalization schemes were em. j played. With appropriate time-step size voldfraction dis.tributlJRs pet voidficctions inferredfrom the measured data. Ncnphysi- 6 col oscillations in spctlal void profiles were observed when i;; - a large time step was used. Comparisons of TRAC 4 predic- _~W tions with results obtained using three codes af the P.ELAP family wre performed. 'i"" d . fMIMW*"*** WIM. hie 5EE$$EEEEEETI E '
*
- ij (,Y"" .
o INTR 000CT10N y 9 . Simulations of two small vessel blowdewn tests were per- - formed using the best est.tmate pressurized water reac:or (PWR) TRAC.PF1 (Version ll.6) computer code.' The pri-rnary objective of these experiments, conducted at9General U " Electric Company (GE) (Refs. 2 and 3), was to investigate blowdown phenomena such as two-phse mixture level swell and to investiga.e void fraction distributions during blow- -- down. The two-phase blowdown phencroeson is a subject of - Suppressen great laterest to both che chemical and power industries. It . Pool , is particularly pertinent to steam water boilers and to pres. Fig.1. Schematic of GE levet swed test. surized and boiling water nuclear reactor systems.The GE swc!) blowdown facility co pressure vessel, a blowdown line containing an orifice, and a suppression tank at atmospbetic conditions. A schematic drawing of the test section is depicted in Fig.1. Detailed tion as the vessel was blown down from its initial state, par-descriptions of the test are presected in Refs. 2 and 3.tially Thefilled w'.th saturated water at ~6.90 MPa (1000 p pressure vessel was a 4.27-m- (14 ft)long and 0.3048-m<l.ft) This study presests two s'unulations of two GE level tes diam vertically oriented cylindrical tank. Instrumentation usics the TRAC PFI/ MODI computer code. Compariso included six differential pressure (DP) cells spaced atbetween equal the TRAC-PFI predictions and the experime
, measurements are also presented.
axial intervals in the vesset. The enessurements obtained from 4 these DP cells were used to infer the void fraction distribu. VOL. 69 JUNE 1983 NUCI. EAR TECHNOLOGY 388
! - - 'y
m (l in.) la diameter. The saturated liquid level - 0.009523 was 3.17 rn (10A ft) at 6.97 MPs (1011 psia). The seco l TRAC MSOtt DEgCRIPTION simulation, a bottom break, i was performed fusfog the sam l he TRAC-PFl/ MODI computer code was developed orince size.This at case was init all-f with 3.06 ml les Alamos National Laboratory saturated tojunctiott calculate liquidthe at 6.93 ther.MPs (100$ psia mal-hydsauti
- i break at time zero.
i transient conditions. TRAC-PFl/ that govern MODI solves the s x con.servation e tutive relationships to model the phenomena ' mass, momentum, and energy exchanse be-ween PRiDicT10Ns the phases. AND COMPARISSN WTTli SATA These relationships play a major role in the code predictJotts. Top Bloendown Test The numerical solution scheme tion equations for two-phase flow is carefully formu l d used to toThe solve ate the conserva-top blowdown test was simulated using TRAC- ~ avoid nonphysical oscilladons. The stability-enhancmg PFl/ MODI. two- The TRAC prediction ci the vessel f pressure I step method used for one<ilmensional transient, flow obtained eliminates using the themateria homoge Fluid system modeling using TRAC-PFI/ break nents, MODIjunction. is is compared against the measured pressu accomplished by constructing modules toindescribe Fig. 3.theThe vari.TRAC predictions hb k usla ous components in the system and configuring the compo-cent modules in a manner that best describestional the system. and one cases. with anone with loss additive no additivecoefficient loss coeffx:ic h of 1.0, are al Component models available in TRAC-PFI/ MODI include presented in Fig. 3. As expected, TRAC underpre PIPE, TEE, $ TEAM GENERATOR, ll t transientPUMP, pressure BREAK, when break FILL,orifi an tem.The uses a 14-cal! basicPIPEmodel used totosimulate component modelloss theThe the vessel,GEicvel coefficient transient 1.0 swe aof void was tesdistn% used. fracdon s don is strongly dep BREAK component to inodeldent andthe onblowdown steam interfacial rnomentum phases. With appropriate orifice, exchange and ditime-step d by between adum size an ary conditions. This mode! is illustrated in Fig. tial 2. detail, transient void fraednn distribudons pre cte Vessel top and bottom break locanons using various blowdown orifice sizes were l i was tests, in this study, one blowdown test at each ocat on considered in the GE level swell simulated using the TRAC-PFI code. Test No.1004.3, a*7 topbreak Break Slas = 03525 mm 1000 - Initiet Miscare Level = 3.17 m initial Pressure = 6A7 MPs g "I l Bresk Outiet - Dets 14 u s= j\ s
--- *ssou,e m = 0 m Pressa IK a f.0)
Pressure (K = 0,0)
- 5
, 700 -{\ .
4 u 7 000 k4
=
1 n 10- ~ e 500 . . N @q-- 9
, \ \g - - 3 g .
7
=2
} 200- \ I 4 \ N 100-3 . j 2 * -CellNumtwr i 300
, e 200 250 So 100 150 0
Time (s) ( 3 Fig. 3. Comparison betwen the comptned an A Fill Component Numper (Mess Flow Rate a 0.0 kg/s) deprescrization of GE levet swu test 1004-3. 389 Fis. 2. TRAC.PFI nodios for the GE smalt sessel blowdown. VOL. 69 JtlNE 1985 NtJC1.EASL TECHNOLOGY
~
rsassan a w v-r rsius m v = uv a n r nu.u.- . . . . . 4' Axial Distance (rni
- Axlel Distance (m) 2 3 4 1
2 3 4 _ i a . 1 e a i a- i t a 40 s gf 1.0 -
. t er 10 s 1.0
!! ~
i l f a Data
- TR AC.PF1 (At = 0.3 s.14 Noce) j 0.8
---- TRAC #F 1 (at e. 0.3 s,28 Nodel lll 0.8 -
TR AC#F1 (At 85 0.1 s,14 Nede) l
~4-
-0 Data h
-*- TR AC.PF1 (At as 0.3 s.14 Nodel l l#
-- A (A M s,28 N el ]
OA [
.f0.5 -
l8 ,! --e- TRACPF1 (at = 0.1 s.14 Noce) ,f y f. - \ m I e 5 ! 3 > > ! g l Ol u I 04 -
) l D.4 1
JL,$ 4
- O 0.2 ,
p 02 =
/ D ./
'''i''r''i 0,2 4 6
f'
8 10 12 14 0 0 2 4 6 8 Axial D: stance (ftl to 12 14 0 Axlal Distance (ft) , Fig. 5. Comparison between the computed and measured vo flg.4. Comparison between the computed and measured void fractions at transient time r- 40 s for ses: 1004-3. fractions at trar.sient time r = 10 s for test 1004 3. TRAC-PFI produced a smooth and accurate predicdon of TRAC-PFI compare favorably with the vold fraction diatd- the spadal void distribution, butions inferred from the mensured data. Predicted and men. sured void fracdon distribudens obtained at 10.40.100,and
- Bottom Blewdows Test 160 s after rupture are oresented in Figs. 4 through 7, respec-tively. Nonphysical oscillations in the spatial void profiles T12e bottom blowdown test was simulated using TRAC-were observed whers a time step of 0.3 s was used. The ese!!- PFI with homogeneous flow friction factors and additive loss lations were eliminated by decreasing the tirce step to 0.1 s. coefficients of 0.1 and 1.0 in the break junction. As shown An increase in the number of computatior:al cefts used in thein Fig. 8, with an addidve toss coefficient value of 0.1.
vessel model from 14 to 28 did not clirninate the oscinadons. TRAC overpredicted the pressure in the transient period of It is suspected that these time-step related oscillations result 20 through 30 s and underpredicted the pressure at transient from the explicit velocity dependence in the friction factor times >30 s. This discrepancy between predicted and men-function forms and the strong void fraction dependence in sured results may be due to the lack of sufficient spadal the interfacial shear corralations, detail near the break to capture the exit void fraction. The The TRAC prediction of the void fraction distribution TRAC predictions of the tweshase level compare favorably at 100 s after rupture is compared with experimental results witt, the ineasured data, as shown in Fig. 9. and predictions obtained using RELAP4/ MOD 6 (Ref. 4), RELAP5/ MOD 1 (Ref. 5), and RELAP5/ MOD (Refs. 6 and 7) in Fig. 6.The RELAF4/ MOD 6 slip model produced CONCu!S10N1 nonphysical oscillations in the spatial void distribution that 1:verely ahered predicted void fractions in the downstream Predictions of two GE level swell tests were performed. cells. The RELAPS/ MOD 1 resuks were also osci!!atory, but The TRAC PFI/ MODI code results compared favorably with not nearly so severcly as those of its predecessor. The new the snessured data when appropriate dme' step and grfd sites interfacial drag model trnplemented in RELAP5/ MOD 2 were used. Nonphysical spatial oscillations in the void frac-improved the spatial void predictions significantly with 27tion distribution were observed when large time steps were axial cells in the vessel.' The codes RELAP5/ MOD 2 and JUNE 1985 VOL. 69 Nuc12AR TECHNOLOOY 390
H:uan TWO-PHASE BLOWDowt* ruun.savna l Axial Distancs (ml li Axial Distance (mi 2 3 4 I 3 4 1 a e 1 2 4 i a 6 i a 1.0 - t a 100 s ID . t a 160 s i , O Dau - f} (
-+-- TR AC# F 1 l I
[ g3 . ~--RELAPSMOD2 lt -
~P- RELAP5/ MOO 1 OE s
~6 -RELAPeeMC00 I U
- Q &u l f
- l
-*- TR AC PP I s' (at W 0.3 s,14 Nodel [
l
- 0.6 - --- TRAC #F 1 j l 0.6 l (at = 03 s. 28 Node) '
)
b 8
-e- TR AC-PP 1
~
E $ - (At ou 0.1 s.14 Nodel fl
> g' i '5 h
I 0.4 *
/ $~w' O.4 -
l I jfb . g W O O.2 - M 0.2 -
.*' s ,
/
0 O 2 4 6 8 Axial Olstance (ft) 10 12 14 02'''''''''''10 0 4 6 8 12 14 Fig. 7. Comparison between the computed and measured void Fig. 6. Comparison between the computed void fractions fractions at transient usingvarious time t = 160 s for coda and test 1004 i = 100 s for test 1006 3.
.7
" % ---- . _ s - 6 800 -
o Experirnental Data
,N.%.%*N, "
6 o
- TR AC Calculation (K
- 0.11 , 4 ,,
,600 -.- TRAC calculation (K - 1.0) g o
^
3 400 - o - 2
- w 1 o
o - 1 200 -
- f O
30 40 10 0 0 Time (s) d l Fig. 5. Compenson of TRAC prediction of venet pressure and experimenta 391 VOL. 69 JUNE 1945 NUCLEAR TECHNOLOGY d
Hassan TWO-PHASE BLOWDOWN PREDICTIONS MRM - 3,o 10 1
- 1. " TRAC Pf!/MODl: An Advanad Bat Estimate Ceewater Program for Pressurized Water Rcactor Thamal.Hydraulie Anal-25 ysis." Los Alamos Nanonal Laboratory (;o be published).
8 ,,,,, E
- 2. J. A. FINDLAY. "BWR Aefil!.Reflood Program Task 4.5-3: ,
24 ~} Modd Qua!irication Task Plat." NUREG/CR-1899. U.S. Nucisar {6 s
.e.o.a.,em. D,o
. s., s R..=m Com su-av.. i9sn.
- 3. B. C. SLIFER and J. E. HENCH.
- Loss of Coolam Accidems e TR AC Calcutstion and Emergency Core Cooling Models for General Electric BoiEng 4
Y~L . 1A g Water Reactors," NEDO-10329. Oscara! Elec.ric Company ( Apr. g - p g 39733' 2 -
. as
- 4. "RELAP4/ MOD 6: A Computer Program for Transient
- Thermal-Hydraulic Ansinis of Nudear Reactors and Related Sys-
. . tems," CDAP TR 003, Idaho Nadenal Engineering Laboratory 25 0 10 1ti 20 0 5 Uan.1978).
Time (s)
- 5. V. H. RANSOM et d., "RELAP5/ MODI Code Marraal,"
NUREO/CR-1526. EGO.2070. U.S. Nudear Regulatory Comrnis-Fig. 9. Compansen of predsted two-chne lswl sad experimen. tal data for GE bottom vesse' : icadown. sion (Nov.1980). 6 V. H. RANSOM st al.. RE!.AP5/ MOD 2 Code Manual." ECC.SAAM4377 EG&O idaho. Inc. (Apr.1984). employed. These oscillations were not in evidence when a CHOW and V. H. RANSOM. "A Simple Interphase Drag
- 7. H.
reduced time-step size was itsed. Reducdon of the spatial Modelgnd for Nitmerical Two-Fluid Modeling of Two-Phase flow aire altered the character of the osdllations but did not Systems." clim-Trans. Am. Nascl. Soc.. 46, 853 (1984). inste thern. 1 l l VOL. 69 JUNE 1981 NUCLEAR TECHNOLOGY 392
Enclosure 4 i Report on Use of Full-Scale UITF Data to Evaluate Scaling of Downcomer and Hot Leg Two Phase Flow Phenomena l I l l l ( l l l l l 1 l l _m ______._ ____ _
i 4 I USE OF FULL-SCALE UPTF DMA TO
- EVALUAili 3EA.Imi 0F NWNCOIER (ECC sTPASS) AIS
! ;un LEs Two-rtA5E Flow PMt.mmenA i P. S. Damerell !
a N. E. Ehrich . K. A. Wolfe i ! { MPR Associates, Inc. ) i
) Abstract !
) The first UPTF Downconter Separate Effects Test and the UPTF Hot Leg' Separate Effects Test provide full-scale data useful , for evaluating scaling effects. The downcomer test showed that subcooled ECC penetration down the downcomer at one steam - . flow was greater than would br. predicted from several j correlations using the largest available subscale data l (1/5 scale by length). This is a favorable result fra a , j licensing standpoint, i.e., actual full-scale performance is ! j better than thought. The multidimensional flow in a large i ! downcomer appears to be a key factor in the better delivery at 4 large scale. The het leg test showed that saturated water ] runback to the vessel in a hot leg under CCFL conditions is
- very close (25%) to that predicted from the largest subscale i tests (1/13 scale by area). This is an encouraging result
from the standpoint, of scaling. Further, this test shows there e is a large margin between typical small break LOCA reflux i condensation conditions and CCFL, and that the major scaled , i small break LOCA scaled integral facilities (PKL, Semiscale, ! ROSA-IV, FLECHT-SEASET) operated within the hot leg CCFL i boundary, even though not necessarily at ideally scaled PWR : ii conditions. Finally, evaluatien of these data show that : i runback of de-entrained water in a hot leg during large break ; LOCA reflood is likely to occur in typical US PWRs, and the ' i data successfully explain the observation of runback in SCTF ! (full-height oval hot leg) and the lack of runback in CCTF (scaled height hot leg).
' Introduction
! Research on the effectiveness of the emergency core cooling system (ECCS)
- in a pressurized water reactor (PWR) has involved a large number of ,
! separate effects and integral tests, essentially all at scaled geometry. The large number of tests have provided useful data for models and correlations of various pheonemena, and for assessment of integrated i i , ! 143 !
I { - computer codes for loss-of-coolant accident (LOCA) evaluation. One of . the residual issues with regard to the accuracy of nuclear power plant i calculations is the uncertainty introduced by calculating at full-scale
- while testing and assessing at subscale.
i One of the major effects of scale is the impact of flow channel size on flow patterns and flow regimes in two-phase flow. Particularly during l l portions of a LOCA in which velocities are lower and gravitational forces
- play a much stronger role, it is known that the size of the flow section
- has a significant- effect on the flow pattern, on the transport and retention of water in key areas, and thus on the occall course of a i transient. The latter portion of a large break LOCA and a maall break LOCA are examples of scenarios where gravitational (and hence size) '
effects are important. ! Recently, separate effects tests in the Upper Plenum Test Facility (UPTF)
- have provided the first full-scale data on two key two-phase flow i
~
scenarios in PWR LOCA evaluation. These UPTF data provide a unique l
- opportunity to evaluate the effect of scaling up to full-scale and to Al so ,
- assess the scale-up capability of analytical and anpirical models.
evaluation of these data provide improved. insight and assurance about
- expected PWR behavior. Accordingly, it is appropriate to evaluate the j data from these tests in this regard and the purpose of this paper is to i describe the results of initial scaling evaluations from these tests.
reactor safety ' The two UPTF tests discussed herein and the overall scenarios to which they relate are as follows: i 1. Downcomer Separate Effects Test - This UPTF test investigated ECC ) delivery /Dypass in the downconer of a PWR. It is related to the ! reactor safety question of how soon and how quickly the vessel j refills with ECC water at the end of the blowdown phase in a large l break LOCA. The key phenomenon is the countercurrent flow ; l limitation (CCFL) in the downcomer (i.e., downflowing water in the l i l face of an upflowing steam / water mixture) which is strongly affected l by condensation and by the multidimensionality of the downcomer. This scenario has long been considered to be scale-dependent. US
- licensing rules '(10 CFR 50 Appendix K) artificially require no ECC l delivery down the downcomer until blowdown is concluded. Scale test
- results from the NRC ECC Bypass Program (up to 1/5 scale by length)
- showed ECC does penetrate, and empirical correlations to quantify 1 penetration were developed in that program. These correlations were 4 generally thought to be conservative if applied to a full-scale i P6R. Accordingly, this UPTF test (which was the first of four j downcomer separate effects tests) helps to accurately quantify full-
]
scale behavior. i 14
9 s ,r : I < This UPTF test investigated ;
- 2. Hot Leg Separate Effects Test --
flows in the hot leg of a PWR. It is [ steam / water countercurrent related to the reactor safety question of how readily the drain-back of water occurs in the hot legs during a small i break 1.0CA (e.g., during reflux condensation cooling) and also to how readily i l de-entrained water might drain back during the reflood portion of a- I large break LOCA. This issue has been previouslyAccordingly, addressed with i the separate effects tests up to 1/13-scale (area). i UPTF data provide the first full-scale glimpse at this phenomenon. : ' l I This report presents brief overviews'of UPTF and of the two tests (all of ' which have been presented elsewhere) and discusses the scaling evaluation j of downcomer and hot leg phenomena. Summary Descriotion of UPTF l ! The UPTF has been previously described (References 1 and 2) and is briefly discussed here with emphasis on the downcomer and hot legs. which is similar to a US. 4-Ioop UPTF simulates a 4-loop German Pieta full-size reactor vessel and piping (four l Westinghouse PWR (Figure 1). ECC can be injected hot legs and four cold legs) are included in UPTF, One , 4
. in the hot and/or cold legs of all four loops, or in the downcon of the l The four steam generators are simulated by containment simulator tank.
four steam / water separators and the four reactor coolant pumps are
- The reactor, vessel j simulated by four passive, adjustable resistances. The core upper plenum internals and top-of-core are full-scale replicas.
l 1 f I 1hnv esi 3.Osummee vased tw Not tse @ hussessand. 3d Dio. eve Womed I. caid Las @ scc.humem masse g, sed tag l I temsase.emshuuhw 4 huipsm
- 98'8 W @ N""** 06"I'd i . Isses.Gammesw w See atediaplanAsg l 56aankumme gc Isy gcm.modownesennene.
y, g,g j g f assea
.m.s
.- g, i.
- ** '* g m w i.
! 5- g2Y , g'**='*2 en e
, to-m4 Qep>
~ p 4_
e1 r 6% o ( j l
\ -
Q ; 1 OVERALL VIEW OF UPTF (FROM REFERii!NCE 4 PIGURE 1 d 145
i l
-r :
~
l is simulated by a steam / water injection system with 193 nozzles, UPTF was one for ! each active fuel assen61y which would be present in a PWR. ' originally designed and as an integral reflood system phases of atest largefacility break covering LOCA; the end-as of-blowdown, refill discussed in this paper it has also proven very useful as a full-scale l separate effects facility covering both large and small break LOCA UPTF can operate at up to 18 bar (260 psia) pressure and phenomena. 220*C (428'F) temperature. The UPTF vessel downcomer (Figure 2) has an inner diameter of 4.370 m i (14.3 ft) and en outer diameter of 4.870 m (16.0 The four ft), giving l 250 m (9.8 in). downcomer skirt to the cold leg centerline is 6.64 m (21.8 ft). 750 nm (29.5 in) cold leg nozzles are spaced around. the downcomer as shown in Figure 2. The lower plenum is 2.48 m (8.14 ft) high from vessel i bottom dead-centgr3 to tht; lower edge of the downcomer skirt and volume of 24.9 m (880 ft ). f ' Westinghouse PMt due to the presence of core anc downcomer simulator lower plenumpiping in the U lower plenum. Table 1 compares UPTF and Combustion Westinghouse configuration with that of typical ) Engineering (CE) US PWRs*. l because these l
- Babcock & Wilcox (B&W) PWRs are not discussed hereist Future UPTF tests are relevant mainly to Westinghouse and CE PWRs.
UPTF tests will cover conditions relevant to B&W PWRs. i i M -'gllf=' ' N 1 L
-- s- ~
* /\ 4, l 4 :
/ /
4 it Ep.
/
d -% s -h
-]
'r s J' l~ C T '.
j Tii.1 p' [! Hf I ' ill7.-.
- r e e-4MW- :s-OAft::i.inI_ "i,_ _ xj
== ,
- 4 :Ml:::r.,
f
',+-x.lf.g g W
~~
g
- ( 7
- 5 - .,, , , s s
! q L ---a= \ l i N#a ,
j
%l
;L my I ....'
I 1 ; ___ ,,, ,,,,,, f
, i g , l{,/__f_i fp 2- ..
g osi.t. 1 ke ms+e f 1i.- o t...ni.n .
, , g ,
3 c.e.e ., a UPTF PRIMARY VESSEL (FROM REFERENCE 2) i FIGURE 2 , 146 _ _ _ _ _ _ _ __ _ ___________________ _ ____ _ ________________________________j
_ . . . .- -_ _. . . .m m.. . . ._ . . . _ . . -
* - TAM.E 1 AttSON F UPTF famefuER Alm i' Gourm==uon unn urtCAL tESTh emen as w-emus m- -sum u;al PIA'S t
Westinehouse CE Ptst l F7F Value volum Pts value Paremeter l 4.87 4.39 4.83 Dounconer CD (15.0) (1464) (15.2) l e (ft) i '! 2s4 258 280 Deescener Sep (10.0) (9.8) (10.2) as(in) 1' 4.64 5.33 6.46 D - amer Neight. (17.51 (11.2) 5ttrt to cold Le9 Center s (ft) (21.8) ! 21.5 24.9 29.7 (1089) (794) ty(ft )glenus volume a (800) ( d has a J -Each UPTF hot leg (Figure 3) is 750 me (29.5 in) inner dia about 8 m (26 ft). A 50' riser section rises 0.91 m (3.0 ft) at the In end the i of the hot leg attached to the steam generator simulator. j horizontal section of hot leg, an internal ECC in.iection pipe There ("Hutze") was no l is located along the bottom edge of the pipe (Figure 4). ' injection through the Hutre in the tests The discussed Hutze blocks in this an report, area of i.e., it is a dead space 2 in the hot leg. A Hutze O.0444 m2 (0.478 ft ), about 10 percent of the total pipe area. Table 2 compares UPTF is present in German PWRs but not in US PWRs. leg configuration with that of typical Westinghouse and CE US PWRs. D TABLE 2 ! COMPARISON OF UPTF MOT LEE CONFIERArt0N WITN TrptCAL 1 utU tm-- Am weu6 son tastatusus tu; rour s i 1 Westinghouse CE Past WPW value 1 Value PtR Vales P;.__;er / 0.790(29.$) 0.737(29) 1.07(48) Diameter. m (in) 1 ! 1.07(4Z)
%dreut te Diameter. e (in) 0.639(!$.2) 0.737 (29) 0.397 (4.25)* 0.427 (4.59) 0.894(9.62)
Flow Area, m2 ggg!) I ] 2 0.441s n! within diesster eines 0.0444 m blected by muta . i ;
2.7a c' 1.25 0 1.2s % $ 1.85 e _ steen g' (4.20ft)
~
(12.4 ft) (4.10 ft) (s.esft)~ generator . Steulator 0.65 m ;l . l l (2.13 ft) i I I +/ D.91 m l a (3.00 ft) 1 = i so i 9
.m . i_
I I / (25.5th). \ ' 4 I i 5 l list Lag
-c Locatten of samme i Domsttemeter seems Vessel j ECC Injeccles j Nea21e j
Core terrol
- tests:
Dese einenstens are for the yw broken isop hetla leg,thewhich intactmes leeps, thethese only het j leg used in the Het Log Separate Effects Test. i ? tue dissestems are stigntly larger (3 86 m and 1.34 m). i UPTF HOT LEG,CONFIGURA110N
^
FIGURE 3 i
==.== .'s.enuma se.e n8 l r
- es.nriw = = . e.n w te.no n 3 l geman p nar = s.mo e is.m as I
^^'
i fA ,' } l W tAl L,,,,*, % mi = , . g ,, ] ,, I e i e n 6. mean
.. so.= =>
l-
- lu al 1
"n i ,
j 1
'~ ' L . ..ui.. ,
i esenow A-A CONFG1 RATION OF INTERNAL ECC IPUECTICM PPE (MUTZE) N UPTF HOT t.EG PImmE 4 148 i
-i'
} Overview of UPTF Downcomer Separate Effects Test The test conditions and results from the first UPTF Downcomer Separate ! and are briefly Effects test are described elsewhere (Reference 3) In . reviewed here. The test was run in two phases: transient and steady. l both phases the loops were blocked at the pump simuistors, and the cold i leg break valve was used to allow flow to discharge from the system.
- Also in both phases e 30*C (86*F) ECC was injected into the three intact A small
' cold legs at a rate of 500 kg/sec/ loop (1100lb/sec/ loop). amount of nitrogen (about 0.15 kg/sec/ loop or 0.33 lb/sec/ loop) was i injected with the ECC to simulate the nitrogen coming out of solution in
- a PlR accumulator. '
4 In the transient phase, the facility was initialized at 18 bar (260 psia) i with the cold leg break valve closed and the containment at 2.5 bar
- (37 psia). The lower plenum was approximately half full of saturated water. The test was initiated by starting ECC floit to the cold legs and l
i opening the break valve to full-open at about the same time. This i produced a depressurization transient with stems (from expansion and i flashing) and entrained water escaping up the downcomer and out the l break, and subcooled ECC water entering the top of the downcomer from the three intact legs (Figure 5). The transient lasted about 25 seconds. e
-B l
- troken /, .- '
! 'I*""' / Cold Lee V, Cind W l l i ^ f1- , l r
- o "
, ,, s N '
sec
/ %
l \ # **I****** j stees riew / s / se=a core o Nve,< ,3 l\ N to aresam j eeeeeeeerisessen / f, ] j
)
t l I /
/ c,,,
' /
f i r i t - , i - j t j
. I h
- ( i ---. et.o. ri 1 =a e., ri ,
/
i i % ~~e ._ # , per stay 11ettr. ) rtasune/ensre tee,ess met t.=,s we.c =%.a
- Due to Depressurtsetion
; OlAWWWA OF Pt.OW CONDITIONS DUfWNS TRANNENT PHASE OF UFTP DOWNCOMER SEPARATE WPECTS TSST PMUIE 5 149
- - - - - - u---
l i j In the steady phase, the facility and containment were initialized There wasat a
' 2.5 bar (37 psia) with the cold leg break valve fully open. To start small initial . saturated water inventory in the lower plenum flowed to the lower plenum, up the was injected in the core whichAfter a few seconds, ECC injection in the j downconer and out the break. The downcomer test was over in about cold legs was initiated (Figure 6).20 seconds at which time the low 4 the core. '
t As discussed in Reference 3, the transient phase showed a mixture Atofthe ECC bypass (out the break) and delivery down the downcomer. i.e., the conclusion of the blowdown the lower plenum was nearly full, i inventory increased during the transient. Local downcomer measurements i showed a strong asyneetry in the flow, with ECC delivery preferentially The steady l occuring on the side of the downcomer away from the break. phase showed nearly comp.lete penetration (about 80 percent) of ECC down ' the downcomer against the upward steam flow. flow Oncewith again, ECClocalpenetration downcomer measurements showed strongly asymetric 4 favoring the side of the downcomer away from the break. ) EM
/
l seek" / cetd Les t +
** $ b spyer _
}
l 1, *= % % Plenius ~r
\ /
j b / % / not , / %g , zasweien ]
- / Y j l m ^ \
' } '
! d It iI 1 , d~' ! ..e
,1
- and Up j i ! j
/ f f+'s)
/
4 Iv t- 1 1
- i i _ _ _ ... ,,.e
---o unter risw
- - nt.,
j
4 por stapitetty,
- seems rien in Cese finittater I Injeetten Pipes
! I ORAMAM OF FLOW CONDITIONS OURN3 PSEUDO--STEADY PHASE OF UPTP DOWNCORAWI SMRATE EPPEC13 TSBT l MOURE8 l
150
t i - i Evaluation of ECC Delivery /Syoass Scaling ' i To best quantify the results of the UPTF downtoner separate effects test for evaluation of scaling, mass balances were perfomed for each test phase. The results of the mass balance are shown in Figure 7 for the
- transient phase and in Figure 8 for the steady phase. '
i Figure 7 snows that when the transient started there was a period during This storage was inferred : which ECC was stored in the intact cold legs. frns thermocouple rakes in one cold leg which showed subcooling appearing .
- There were no direct measurements
- at all locations over this time frame.
I of the amount of mass stored; the curve shownVessel assumes the closed and inventory decreased pipes filled according to the injection rate. i slightly while the legs were filling due to flashing. When the cold legs filled and water was being delivered to the downcomer. Small vessel invertory rapidly increased, indicating ECC delivery. l indications of bypass out the broken cold leg first appeared at this time as well. Over The " a period of about 15 seconds, delivery and bypass both spike" in delivery is apparently attributable to a brief
- occurred.
i septying of the cold leg inventory above.
-- a corresponding de At the end of the rather than measured, as discussed i i
depressurization (about 25 seconds), the lower plenum was essentially i full and less than half the injected water had been bypassed out the l ! broken cold leg. ( tol ICel , l i ' e.== r,' ! "4 "t.PM:;'{try
===
0 8 kl
)
_..f.uwu.=.w.u L___. ._,
'~ - /
,,e r ~, aY /
= trute. (s' d tog Fles y
i ,7 7 a o i u a es 1ese efter Start of NC IWasst.e (ses) [} LFTF DoWNCoMeR SEPARATE sFFECTs Te8T MASS eAL.ANoe FoR TRANelENT PMAse neure 1 ~ 151
-- ,--- r--.- - , , - . - , - - -- . . . . . - - . . .
,r l
Figure 8 shows that duringfilled, thefollowed steadyby.test there substantial ld leg. At was also period during which the cold legs delivery to the vessel with limited bypass out the broken coof ECC in approximately . 20 seconds after the startand only 20 percent of the ECC had b plenum was essentially full, bypassed out the broken cold leg. they are The evaluation of scaling using the UPTF te Figure 9 shows a downcomer dimension discussed - first below. plot using the parametersgj
- and jf* where g ((p f . p g) g y)1/2 j*
g
= M g bg)1/2 7p g g
j f* = M f (P f)1/2 4 { g ((pf _g p ) g )1/2 M = mass flow rate of gas or liquid where 2 or 39 ft 2 forUPTF) A = downconer area (3.62 m p = density of gas or liquid f g = gravity W = downcomer circumference (14.5 m or 47 i 1
/
. === ( l M ,les
. ..t.. ~
i + y ("',',F:"lll!.'!".'. ; i'~~ihn k .f.r / [ f u.} f w. ic7_ he oms em j .
/
~
tten after geseg of E tedertie (ser! l UPw oowwooman surAnm arrects raer uAss nauwas con ersAoy Pnass PleunE 8 152
a 1/2 , j, i Correlation 1: (j g* - F j ,*g Tcond)1/2 + g jf '
~ urn ASSOCETES 1/2 + gjf el/2 = C j
p.rs-as-isce stastsy Correlation 2: gj *1/2 . y j i, i ," I* g ) d*" f dg,* Tcond " de in J Musente Flux Scaling J
- Scaling Correlation 2 Correlation 1 Correlation 2 Correlation 1 Parnanter --
4 0.209 0.281 O.209 F 0.281 f 0.822 0.896 0.822 M 0.896
) 0.230 0.369 0.344 0.250 C
i I 0.20 4 w
~
I i i \g UPTF Result 0.15 - \
\ @ UPTF CCFL Prediction Based
* \
l
\ on Two Correlations from 5 scaled Tests J* Scalinc) l j j UPTF CCFL Prediction Based
- Ns 1 on Two Correlations from-l - g l
5 0.10-
* % % scaled Tests (Momentum Flux 4
C g 2 [N g/ %
% ' Scaling) l
! g s% % * % ' % L ,' i e 1
**=.
3 l 'A f f 0.06 ' N-2 i 5 i l E i E s % l g 0.06 0.08 0.02 0.04 O Dimensionless Liquid Flow Penetrating Down the Downconer, j,
- 1
- COMPARISON OF UPTF DOWNCOMER TEST RESULT WITH PREDICTIONS BASED ON 1/5 SCALE TESTS FIGURE 9 153 _
f
,r J
j From previous scaled tests in the NRC ECC Bypass Program (Referen - 8), j* correlations were developed using data'fromA convenient sumar ; J J 1/5 scale (by. length). Reference 4. l ; l The curves shown on Figure g represent the CCFL boundary calculated ; UPTF based on the largest scale data available from th: tests (1/5-scale). the two lower curves represent a " constant calculating the UPTF boundary: momentum flux" scaling approach with two different forms of correlating - the 1/5-scale data; and the two upper curves represent a " constant j' I scaling approach with two different forms of correlat i data. correlated some of the subscale data more favorably -- there is no clear ! ! The lower curves are the NRC-basis for reconnending one over the other. i recomended approach for downcomer CCFL based on the scaled tests !
- The upper curves represented a more " realistic" NAC ECC Bypass Program. ,
approach which was not reconnended by the RConbecause it could n i ! to be conservative, at full-scale based the scaled demonstrated i tests. The main result of the UPTF downcomer separate effects test is l that the full-scale test shows more ECC penetration than would be j predicted by either the NRC-recommended or realistic approach scale. Hence, there appears to be acondensation, beneficial effect of large to large scale. channel l which may be related to improved The observation of the strong asymmetry in the ! hydraulics, downcomer, or to both.i.e., preferential ECC downflow on the side away from break (see Reference 3), indicates that the large channel effect is ' l probably significant. i i The min result of the transient phase system is of t'hetodown continuing even while the primary Iower plenum i depressurize. Although scaled tests suggested this The would occur, this UPTF test was !- full-scale test provides the best direct evidence.with regard to reasonably PWR-typical thelower rate somewhat low and l subcooling. The ECC injection was l depressurization somewhat prolonged in comparison to a typical PW l j but these differences do not affect the validity of the o discussed above. benchmark analysis case for computer codes. . i It is not feasible to run in UPTF a direct counterpart transient test to ! previous scaled ECC bypass tests, due to some particular choices made in plenum volume and containment pressure in the j PWR-typical) Accordingly, futura downcomer separate previous sca led 4 facilities. effects tests will focus on steady-state downcomer CCFL conditions, in an attempt to further evaluate scaling by comparing UPTF results with CC
- curves derived from previous scaled tests. I 154
l i i
- j
. Overview of the Hot Leg Separate Effects Test j The test conditions and results from the UPTF Hot Leg Separate Effects
- Test are described elsewhere (Reference 9) and are bri efly reviewed
- here. The test was run using only the broken loop hot leg of the UPTF.
l The test was performed as several ste.My phases, each consisting of staan . injection into the primary vessel unwh flowed out the broken loop hot l 1eg, and saturated water injection in the steam generator simulator i plenum which could either flow back dawn the het leg toward the vessel or i out of the systes through the steam generator simulator (Figure 10). Six separate steady flows were obtained at 3 bar (44 psia) system pressure ~ , and 10 flows were obtained at 15 bar (218 psia) systaru pressure. In all ! cases water flow was established prior to steam flow. The intent of i obtaining several flows at each pressure was to " map out" the CCFL ! boundary. Also, one of the flows at 15 bar . simulated conditions in a i Westinghouse 4-loop PWR during the reflux condensation mode, which can occur during an SBLOCA. l , i i d l Seston A.A SeadenB4 , s, n.m 4 ! I h Ans -emas mi
'# oom sur een= 43fM ml Nem \
l + __. . _ _ . 1 ( __. Q ! % C 7Nm ="
- 1 - -s I
$tsamhuster asseman i
4 sL 'g,W: > - 1, l i % = esth6sessa M l A- hostpuum i amme ning 8h'" 11 / s i tem
% sgehu 4M * .
LY L ) n le 2.secess/8"Ima i=P esse aseuseg a 1 luthe J i ) i UPTF HOT LES SEPARATE EFFECTS TEST i OVERALL PLOW CONOmONS wnohi nsPERENCE 41 f FIGURE 10 i n i 155
l
= #
-. e i.. w in -
a w nm c 5.- i I
.i - %7.2 :
i a 8 s 1
~
i , ?,.$**d5-*Y E n, l
. sb./f"* n, 4
- m. Hm. s. er - _n: .m r. I't 4.,vsa urir war tea sepAnAn erween nur SuheedARf 0F DATA RGURE 11 Figure 11 shows the measured flows at the two test pressures, and Figure 12 shows the data on a dimensionless j* plot, where j* a M g(P)1/2 g /p 9A (p f -p g)gD) h g
t jf* = Mr(O)1/2 f /of A (@ f -p g) gDh The variables are as defined previously and Dh is the hydraulic diameter. ' which is .639 m (2.10 ft) for the UPTF het leg at the "Hutze". r 4 6 ::Y.'::".M/ l : 1 ,7-i -
%'."l"; 5:'"
.: u. .
. . a.. ir-'*
}u- p. .. g- -
.3- w, ,,
o ::s' g-
.J. 77
~
.t ..ss .. s.ts s i
em
.m
- .g.
WPT,anguLTS H0f LOSs.0usW 99.pemAN OseJ'BPPECT5 PLAT TEST PegamE W W
i
-E 1
On _ the j* plot, the 3 bar (44 psia) and 15 bar (218 psf a) data correlate t favorably. The line drawn through the data on Figure 12 is the "best-fit" experimental correlation to the UPTF data. l l l i The -results of this test provided direct demonstration that there is significant margin against hot leg CCFL during the reflux condensation
- phase of an $8LOCA. This is shown in Figure 12 by the fact that the i
" typical" point is substantially below the CCFL boundary. This point was chosen based on conservative assumptions such as relatively high power and one steam generator inactive, etc. Accordingly, this result provides q direct and convincing evidence that substantial margin exists.
j Figure 13 shows the measured hot leg level and void fraction for all of i the tests, plotted against j
- the dimensionless gas . flow. Thes'e data i are from a three-beam gassaa 8en,sitorneter located just on the vessel side i
' of the hot leg riser bend, as shown on the figure. There is no "Hutze" obstructing the bottom of the hot . leg in this short section of hot leg. l The data clearly indicated a stratified regism and show significant water presence in this region of the hot leg. These data appear to show that CCFL is being controlled by the hot leg (i.e., CCFL is not occurring in the riser or steam generator simulator), since water is not absent from l the hot' leg when there is zero net penet' ration to the vessel. i pZ .04 lJ.*rr C.1;h - m
-p_ ...
i w._ LAs - . ! g( , t ' \. -
,\ "-
us-m.i::-- ". l A, r.,r - = ="a- ; ! I a. rw er 4 menan u-
"e J gw I I i ; ; .. . - - - -
- y
- m.
g ,,. . . . - - .
- y . . -.
8 d he$ e 4
~
.. j' "
LFTF Hof leo SEPAllATE EFPECTs TEST DAEAgustED Mof LEO LEVEL AND VOC PflACTION AS l A FuMCTION oF oWENSIONLK85 STEAM VELotNY l . u. I 157
i J f Evaluation of Hot Leg CCFL ' Scaling fII Several theoretical and scaled, separate effects studies of hot leg CCFL l ' or generalized horizontal channel CCFL have been carried out, including: L
- Richter, et al (Reference 10) - 1/13 scale by area compared to Westinghouse PM l
- Gardner (Reference 11) -- Theoretical i
.0254 m (1-inch) square channel i
~
- Wallis (Reference 12) --
l (approx.1/660 scale by area compared to Westinghouse PWR) i - ohnuki (Reference 13) -- 1/840 scale to 1/93 scale by area compared to Westinghouse PW i
- 1/210 scale by area compared to
- Krolewski (Reference 14)
Idestinghouse PE l i Also. Transient Reactor Analysis Code (TRAC) predictions of the UPTF test were perfonned. Each of these previous studies provides a way to predict full-scale hot leg CCFL behavior. In all cases, j* is the key parameter l in scaling. 1 ! Figure 14 shows the UPTF data compared to the full-scale predictions In the i based on five of the six studies mentioned above, on a j* plot.
\N (Mn
- \,
l Q-ax /=--, _ i 4 pv yN'n2 = i ,, useem=> j
==
l >
- o -
T x&% o ne
', . m. o. o.
- a ww n .swa === T r OERdran00N OP DaBL TO TIEENIETIGhL tegenLB Age COINIE.Anoud8 FRome meALL SGALE TasTB FlulNW to 158
t g [." im%a U*E7 d=% i 4 1 M."-r-j [/ / dg s.
, w
- s,*. m.7 toimee
"* 8 f,,',,",,,", w" "'""
N -
- N er (a.r .c
- ::. -
n. 3 Q. ,,
- _, . m, ,
- we - . .aue . v.
- u. '
, gew _
I u. leur i s.e . e . : e we . .
. . en -,y. ,,
s.3 L u. I j
". 4 a s .:.
sammensais- e--nc.: se . u n,. s,. UPTP HOT LEG SEPARATE EFFECTS TEST i COMPARtSON OF UPTF HOT LEG VOC PRACTMS TO ! 3 WALLS CORMELATM rnURE ils i case of the Wallis correlation, which is a [*/ void fraction correlation, i the comparison is on Figure 15. The results of the comparisons shown on Figures 14 and 15 are as follows: i - Very close ag/eement is obtained between the UPTF data and the Richter, et al correlation, which is the largest subscale data i previously available. The agreement is 15 percent. This agreement confirms that the j* correlation approach appears to be valid. The
- close comparison indicates that scaling up across an order of
! magnitude (ba. sed on pipe area) is successful and is therefore. an i encouraging result. Closa agreement is obtained between the UPTF data and the Wallis correlation which is based on void fraction rather than liquid I flow. This indicates that the basic approach of this correlation ' (once again, a j* correlation) appears correct for scaling, but that implementing this model to calculate liquid flows is dependent on
- knowing an accurate void fraction.
4 i 4 159 _ . d -.
Significant Krolewski deviation is observed in the case of the - Ohnukt correlations. and scaleinofthethe underlying This is considered to be due to the sea 11 bend previous tests. tests and the strong effect of the riser l i angle) could significantly affect. .,the this sensitivit length flooding.th and up" these smally is PW-typical; It is nottoknown this makes it difficult " scale-if I scale results. favorably with the UPTF data.The predictions using the Ga i the model (f.e., unstable stationary disturbance) does no 1 i - realistically reflect the true flow behavior in a PW hot leg. r to The predictions changing gas from TRAC show a nearly "bi-stable" behavio l flow rather than the reasons are still being investigated. gradual CCFL boundary. The [ Overall, the comparison with previous theoretical and scaled s-res very favorable in that the results from simulated hot leg separate I accurstely predict full-scale behavior. effects tests with one o i { In addition several to these Plat integral tests separate of small effects comparisons discussed abo conducted.
- In the small break case, these faci' ities demonstrated refluxand larije i,
condensation occurs without apparent hold-up due to hot leg CCFL. major small break facilities investigating reflux' condensation The are: Semiscale (References 15 and 16) - 1/1705 scale s FLEcliT-SEASET (Reference 17) -- 1/307 scale L - 1 PKL (References 18 and ig) -- 1/134 scale ROSA-IY LSTF (Reference 20) -- 1/48 scale 1 The conditions achieved in reflux condensationcorrela SBLOCA UPTF results facilities in Figure 16. are plotted on a j* graph along with calethe tests i conditions" which of i roughly Also shown in this figure is a band of "PE envelope conditions. This SBLOCA reflux condensation figure shows that although the scaled facility i the CCFL boundary, conditions tend to be scattered about the gr ' as are the PWR conditions. deviate most froe PW conditions, tend toThe be PKL points, which a result of area eg the hot l i scaling other tests.used in these tests, which did not seek to preserve j* a I The major conclusions, though, are that for all of the 4
+
I 1 4 i i - 160 3
l l l 9.#
- yhes.,
s 1- } {b== j
.au , ..m 3
j ,., y .,,,,,.,, S.8 A DEA-IV4.Str
,- i
.:;L.. 1, 't._ "
. .,,$g;;r ,
cowasou or meAu.-sens racun i fisFLAN CONDetEAfl006 10 uPW TEST RESULTE EXPEIWMENTAL CONDITIOett '
' PeguftE te i
i facilities. the observation of reflux condensation without hold u - het leg CCFL facilities did not is consistent distort PWR with hot the UPTF data, and that the sc rom phenomenological way. leg behavior in a major The reflux condensation results are applied to USe PWR's 17. on F This Hgure Westinghouse (1160 psia). and shows CE plants hot(3800MWt) leg CCFL curves calculated for the m condensation Also shown are conserva(tively in bothcalculatedcases) at 80 bar both cases.. conditions for both plants. SBLOCA reflux The large margin is evident in In thethe during large break reflood phase case, of thehot leg CCFL transient., is only an important on cons The major, large scale reflood facilities which allow a detailed are: evaluat%n of hot leg eff Cylindrical Core Test Facility (CCTF) - 1/21 scale i Slab leg Core Test Facility (SCTF) -- 1/21 scale with full-height ho 161
.35
.H-
- .o- -
! .r .n -
.n . i,s =_ . = .,
an. n ,=4 .. g g... ,.g =
- i. .=i =.
- 6. -
j 2 .- lg I*I *y s it [J . D=-
=-
- !,3::M -n=T3 i
g-n . - .i l - A =. . <.L . s 1..' " .' J ,,,_ i
..e l
j ' g3 c lli'aSU" ir*="'nalL j
~' *i j =- '_ 'gg .,
___C.ne,tt.as %._ Canditi.e.h j* ! 1 _(_.__ 4 fin!#uf"FJ :J'J" l i ai -
**'4:#::* 4!!ia": W.M::,,
l
's l
1 s 4 - s 4 s s mass n nae. er Cem mne u.ser, l Ar (kg/s) la One N.t Lag ). i 1 l PnEDICTED HOT LEG GCPL BEHAVIOM IN U.S. PWRs ! { COMPARED WITH 85LOCA REPLUX CONDENSATION { PLOW CONDmONS i Pieuna w . l- , l l i In CCTF, no evidence of counterflowing water during reflood was observed. [ 1.e., any water reaching the hot legs tended to be swept through the l l- prieary coolant loops by the steam flow. In SCTF, though, hot leg water l i runback to the vessel by countercurrent flow was observed and directly ' ] measured. It is noted that due to the unique cross-section of the SCTF l i hot leg -(oval) there may have been greater de-entraineent and more water l l available for runback in the hot leg. The conditions achieved in the CCTF and SCTF 1arge break reflood f.ests l j arc indicated on a j* graph along with the correlation of UPTF resdits in Figure 18. The steam flow j
- associated with typical PWR reflood l conditions is also shown on tEis graph. As. indicated on the graph, i I
counterflowing water during reflood would be expected in SCTF but not in
- CCTF, i.e., consistent with observations. {
The CCTF/SCTF difference is i due to the height of the hot leg (full-scale _ in SCTF but not in CCTF). ! The figure shows the SCTF results, in this regard, are closer to PWR-l typical. 1s Figure 18 shows, counterflowing water would be expected in l 4 both Westinghouse and CE P6R's. ! i
; 2 4
i
; l 162 :
4 u ._ . om.e ree us.a u ..o _ __.L . . a
,,, :D EiEU?ah"** "" ""'"
i u. ua '
! i i
E u-u-
\-
% h::2! E P
~ '
u- I b"MM="Y I I
""- ~-
u-l r?"c':::ct'.".t" usu e an ins === a====# l u.-_ r ! l
. s.n .
4 t e-
' 'a l l us s.u i,,
W i i svaLuanon or arr Lee numaAon ouses LAnse ansAu tocA neptoon we ccw. sow 1 Alep A PWR Basm ose LMF mot Lee Tas7 Result 5 nouns s
\
Conclusion ., l 2 The UTPF Downcomer separate Effects Test andects Test both have Hot Leg S
- tests provided the direct useful information for evaluation of scaling conclusions t results convey favorable . For and downcomer or, a .e., water penetrates to the reactor vessel encouraging l
subscale results. hot Ing as well as or better than wouldthrough a be predicted from i For the downcomer situation, the present test data do i e notfrom up provide suggest previousatests. broad enough base to evaluate the gaccuracy CCFL of sc j that j* sealingThe UPTF results presently available though, do ' conservative approach, andfrom that previous scales provioes at least a approach will have to await upcoming UPTF test resultsdetermination o the UPTFetdata (Richter, al show that predictions from the largest In the hot leg, . (*5 percent). at 1/13 scale based an area) are subscale quite tests accurate based on the j* The parameter, correlation which gives this successIul scaling is i approach. indicating I that: (1) Application during $5LOCA af the UPTF het leg results to US PWt's indic reflux ' margin between condensation, (2) during large break LOCA reflood runbackactual as expected; flows and and the entrained in the hot legs. is likely for water de- \ \ f_______.-_-------' ._ .
l i i References 1.. Hofmann, K.,
" Status of the German UPTF Program," presented at the 13th Water Reactor Safety Information Meeting, October '22 - 25, 1985.
- 2. Weiss, P., Sawitzki M.
Loss-of-Coolant Accidentand Winkler, F. ; "UPTF, a Full-Scale PWR Vol .49,1986. Program," Atomkern-Energie Kerntechnik, 3. Hertlein, R. and Weiss, .P.; "UPTF Experiment: PWR ECC Downcomer Countercurrent Flow Under Steam and Two-Phase Upflow Condition," presented at the October 26 - 29, 1987.15th Water Reactor Safety Information Meeting,
- 4. "1/5-Scale Countercurrent NUREG/CR-2106, November,1981. Flow Data Presentation and Discussion,"
5.
" Analysis of ECC Bypass Data,", NUREG-0573, July,1979.
- 6. " Application of Batte11e's Mechanistic Model to Lower Plenum Refill," NUREG/CR-2030 March.1981.
- 7. " Analysis of Flashing Transient Effects During Refill,"
NUREG/CR-1765, March,1981.
- 8. "Sunnary of Refill Effects Studies with Flashing and ECC Interactions," NUREG/CR-2058, November,1981.
- 9. Weiss, P. A. and Hertlein, R. J.; "UPTF Test Results - First 3 Separate Effects Tests," presented at.the 14th Water Reactor Safety Information Meeting, October 27 - 31, 1986.
- 10. Richter, Horst J.; Wallis, Graham B; Carter, Kelly H. and MJrphy, Stephen L.; "Deentrainment and Countercurrent Air-water Flow in Model PWR Hot Leg," Thayer School of Engineering, September,1978
- 11. Gardner, G. C.,
" Flooded Countercurrent Two-phase Flow in Horizontal Tubes and Channels," Int. J. Multiphase Flow, Vol. 9 No. 4,1983, pp. 367 through 382.
- 12. Wallis, G. B.,
College Report No. " Flooding in Stratified Gas-liquid Flow," Dartmouth 27327-9, August, 1970. EB
I l
- 13. Ohnuki, A.. !
" Experimental Study of Countercurrent Two - phas e Flow in Horizontal Tube connected to Inclined Riser."zszJo 4
! Science and Technology, March, 1986, _ pp. 219 througn uclear i i
- 14. Krolewski, S. M., .
" Flooding Limits in a Simulat <
Hot Leg," Massachusetts Institute of Techaology, ed Nuclear reac of Requirement for a B.Sc. (1980). i Submission as Part ; i 15.
- " Experiment Data Report i Tests S-NC-28, Idaho, December 1981.S-NC-3, and SNC-48," NUREG/CR-2454for Circulation S 1
, prepared by EG8G
' 16. " Experiment Data Report ! for Semiscale Mod-2A Natural Circulation 1 Tests 1982. January S-NC-5 and S-NC-6," NUREG/CR-2501, prepared o, b 17. ' Condensation " NUREG/CR-3654"PWt FLECHT-SEASET e ux S Westinghouse, Electric Corpora, tion, August 1984.EPRI NP-3497, W
- 18. Mandl, R. M.,
i and Weiss, P. A., Mechanisms during Small-break "PKL Tests on Energy Transfer . No. 2. March-April 1982. ' LOCAs," Nuclear Safety, Vol. 23, ~ i
- 19. Thompson, S. L., Kmetyk, L.
Circulation Tests," preparedN.by, "RELAP5 Sandia Assessment: National PKL Natural i NURES/CR-3100. SAND 82-2902, January 1983. Laboratories,
- 20. Tasaka, K. et. al .
- Natural Ci*rculation, tests at"The Results of 5% Small-Break LOCA Tests an 1986, NUREG/CP-0082, Volume 4. Fourteenth , October 27-31,Water Re 4
k a i . j i 1 1 i i I . i J MS
Enclosure 2 Verification Analyses for Vold Fraction Prediction 1 i l i
]
p I i I l I}}